Dr. Mark Paradis

← Student Resources · Last updated: August 2026

Reports and arguments

The most common problem I see in undergraduate papers is not bad writing. It is a well-written report submitted where an argument was required. A report gathers what is known and lays it out. An argument takes a position on a question that has more than one defensible answer, and then earns that position.

The test is simple: could an informed person disagree with my thesis? If no, it is a fact, not a thesis. "NATO expanded eastward after 1997" is a fact. "NATO's eastward expansion was the principal driver of Russian revisionism after 2008" is a thesis — contestable, and therefore worth defending.

Kinds of papers you will be assigned

• Response or reaction paper. Short. Engages a specific reading or set of readings. You are not summarising — you are identifying the argument, locating its weakest joint, and saying something about it.
• Literature review. Organised by debate, not by author. The structure should tell the reader where scholars disagree and why, and it should end by identifying what remains unresolved.
• Research design. A question, a theory that answers it, testable hypotheses, and a plan for evidence — cases, data, method, and what would count as disconfirmation. No results, because you have not collected them yet.
• Empirical paper. The full arc: question, literature, theory, hypotheses, evidence, analysis, and what it means. This is what the research methods sequence builds toward.
• Case study or policy analysis. Applies a theoretical framework to a specific case or decision, usually with a comparative or evaluative payoff.

Starting from a question, not a topic

"Populism" is a topic. "Why did populist parties gain more ground in Central Europe than in Iberia after 2010?" is a question. Topics produce reports; questions produce arguments.

A workable question is:

• Empirical, not normative — unless the assignment is explicitly normative. "Why did X happen" or "what explains variation in Y," rather than "should the government do Z."
• Answerable with evidence you can actually obtain in the time you have, with the sources available to you.
• Narrow enough to defend. "Why do wars happen?" is a career. "Why did the 1969 Football War escalate when comparable border disputes in the region did not?" is a term paper.
• Puzzling. The best questions have the shape "we would expect A, but we observe B." Puzzles give your reader a reason to keep reading.

Building the thesis

Your thesis is the one-sentence answer to your question, plus a compressed indication of why. It belongs at the end of your introduction, and it should be specific enough that a reader could argue with it before reading the rest.

The second version tells the reader what you think, what you are arguing against, and roughly how the paper will proceed. It also commits you — which is the point. A thesis you cannot fail to prove is not doing any work.

Structure

Most of my assignments fit this shape. Adapt it; do not follow it mechanically.

Introduction (about 10% of the paper)

Four moves, in order: establish that the puzzle matters, state the question, state your thesis, and map the paper in a sentence or two. Do not open with a definition from a dictionary, and do not open with "since the dawn of civilisation." Open with the puzzle.

Literature review

Organised by position, not by author. You are showing the reader the state of a debate and clearing a space for your own contribution. Every paragraph should implicitly answer "why am I telling you about this study?" — and the answer is always some version of "because it gets us closer to, or stands in the way of, my argument." Never write a paragraph per source in the order you happened to read them.

Theory or argument

Your causal story, spelled out. What is the mechanism? Not just "X is associated with Y," but how X produces Y — through what actors, incentives, and steps. State what your argument implies that we should observe if it is right, and what we should observe if it is wrong.

Methods

Not every paper needs one — check your assignment. In an empirical paper or a research design, the methods section tells the reader how you are going to find out: what your cases or data are and why you chose them, how you are measuring your key concepts, what analytic technique you are using, and what evidence would count against your argument. It goes between the theory and the evidence.

The standard is reproducibility: could a competent reader repeat what you did and get what you got? That means naming your data source and version, saying how many cases you have and over what period, and being explicit about anything you excluded and why. Case selection needs the same treatment — "I chose Hungary and Poland" is not a method; "I chose Hungary and Poland because they are most-similar cases that diverge on the outcome of interest" is.

In a short response paper, a literature review, or a theoretical essay, there is nothing to put here and you should leave the section out rather than padding it. Where it is required, the assignment sheet will say so.

Evidence and analysis

Where you earn the thesis. Present the evidence, and then do the work of interpreting it — evidence never speaks for itself. If you have a quotation, a table, or a figure, tell the reader what to see in it and why it supports your claim. A table with no interpretation is not evidence; it is furniture.

Alternative explanations

Take the strongest rival account, state it in a form its proponents would accept, and show why yours accounts for the evidence better. Students routinely skip this section, and it is where the best papers are made. Straw-manning the alternative is worse than not raising it at all.

Conclusion (about 10%)

Restate what you argued and what it now rests on, then say what follows: implications for the wider debate, for policy where relevant, and what your argument cannot explain. Acknowledging a real limitation makes you look like a careful researcher, not a weak one. Never introduce new evidence here.

Sections and headings

Use section headings in anything longer than about four pages. They are not decoration and they are not a sign that the writing is too simple — every journal article in the discipline uses them. They force you to decide what each part of the paper is doing, and they let a reader who is looking for your methods find your methods.

Three rules for headings:

• Make them informative rather than generic where you can. "Why Elite Competition Matters More Than Mass Opinion" tells the reader something; "Analysis" does not. Generic labels are fine for the conventional sections of an empirical paper — Methods, Results — because readers expect them there. Everywhere else, say what the section argues.
• Keep the levels shallow. Main sections, and subsections only if a section genuinely has two or more distinct parts. If you are reaching for a third level in an undergraduate paper, the structure has gone wrong.
• Do not let a heading do a sentence's work. The prose has to stand on its own: a section should open by saying what it is about, not rely on the heading to have said it. A reader should be able to remove every heading and still follow the paper.

Headings also give you a fast structural diagnostic. Write out just your headings, in order, and read them as a list. That list is your outline. If it does not describe a coherent progression from question to answer — if two sections are doing the same job, or if the argument arrives before the evidence — you have found a structural problem while it is still cheap to fix.

Where the assignment specifies a structure, follow it exactly. Where it does not, the sections in the previous part of this guide are a reasonable default.

Paragraphs

Most paragraph problems disappear if you follow one rule: claim, evidence, analysis, link. Open with a sentence stating the claim this paragraph will establish. Give the evidence. Explain how the evidence supports the claim — this is the sentence students most often omit, and it is the one carrying the argument. Then connect to what comes next.

One paragraph, one job

The single most common formatting problem I see is the paragraph that runs for most of a page. Sometimes it runs for more than a page. When that happens it is almost never a paragraph — it is three or four paragraphs that have not been separated yet, and the reader has to do the work of separating them.

A useful working range is roughly 100 to 200 words, or four to eight sentences. This is a guideline and not a rule; a paragraph can be shorter for emphasis, and occasionally a genuinely single idea needs more room. But if a paragraph is filling half a page, treat that as a signal to look at it, and if it is filling a full page, it is wrong.

The test is not length, though — it is unity. A paragraph makes one claim. If you cannot state in a single sentence what a given paragraph establishes, it is doing more than one job and needs to be split. Ask of each paragraph: what is the one thing a reader should take from this? If the answer has an "and" in it, you have found your break point.

Reliable signs a paragraph needs to be split:

• You changed subject, case, time period, or author partway through.
• It contains more than one topic sentence — a second sentence that could plausibly open a paragraph of its own.
• It moves from describing something to evaluating it. Those are usually two paragraphs.
• You wrote "additionally," "moreover," or "furthermore" somewhere in the middle. Those words often mark the seam where a new paragraph should start.
• You cannot summarise it in one sentence without using "and."

The opposite failure exists too, though it is rarer: a page of three-sentence paragraphs reads as choppy and suggests that no idea has been developed. If a paragraph is only two sentences, ask whether the claim has actually been supported, or whether it belongs with the paragraph before or after it.

Why this matters more than it looks like it should: paragraph breaks are how a reader knows where one idea ends and the next begins. A wall of text hides your structure. Your argument might be perfectly sound inside that wall — but if I have to reconstruct it while reading, some of it will not survive the trip. Which brings us to editing.

Between paragraphs

Paragraphs need to connect, but connection is not the same as a transition word. The strongest transitions are made by the content of the topic sentence: it refers back to what was just established and moves forward. "If elite competition rather than mass opinion does the disciplining, the timing of the reversals becomes puzzling" links backward and forward at once, without a "however" anywhere in sight.

A diagnostic that takes two minutes and is worth more than any other revision technique: read only the first sentence of every paragraph, in order. You should hear the skeleton of your argument. If instead you hear a list of topics, your paper is a report.

The mistakes that cost the most marks

• Summary in place of analysis. Retelling what a source said is not engaging with it. Ask what it gets wrong, what it assumes, and what it lets you conclude.
• Quotation as argument. A quotation is evidence for a claim you make, never the claim itself. If a quotation is longer than two lines, ask whether a paraphrase would serve better; usually it would.
• Undefined concepts. Democracy, populism, security, power, and legitimacy all mean several incompatible things. Say which one you mean, cite the source of your definition, and use it consistently.
• Correlation asserted as causation. If you claim X caused Y, give a mechanism, and consider whether Y might have caused X or whether something else produced both.
• Cherry-picked cases. Selecting only cases that confirm your thesis proves nothing. Say why you chose your cases, and address the ones that fit badly.
• Rhetorical questions. "But is democracy really the answer?" Answer it or cut it.
• Unsourced claims about the world. Anything a reader could reasonably doubt needs a citation. See the citation guide.
• Vagueness as hedging. "Many scholars argue" — which scholars? "Some evidence suggests" — whose evidence? Precision is not the same as overclaiming.

Style

• Prefer the active voice. "Cabinet rejected the proposal" beats "the proposal was rejected by cabinet," and it forces you to name who acted.
• Hedge accurately, not defensively. Say "the evidence suggests" when it suggests and "the evidence shows" when it shows. Stacking hedges — "it could arguably perhaps be suggested" — reads as having nothing to say.
• Use the first person sparingly but do not contort your sentences to avoid it. "I argue that" is standard in the discipline.
• Define acronyms once, at first use.
• Write short sentences when the idea is complicated. Long sentences are for simple ideas.
• Cut throat-clearing. "It is important to note that" can almost always be deleted with no loss.

Editing for clarity

This is the highest-return hour you will spend on any paper. Not because polish impresses anyone, but because unclear writing actively conceals work you have already done. Students who go from a B to an A in my courses very often do it here rather than by having better ideas.

Editing for clarity is a separate pass from revising for argument, and it comes second. There is no point perfecting sentences in a section you are going to cut. Once the structure is settled, work through these:

• Read it aloud. Nothing else finds as many problems as quickly. Where you stumble, run out of breath, or have to re-read a sentence to get its sense — that is where your reader will stumble too. Mark those spots and fix them.
• Fix the pronouns. "This shows that…" — this what? Vague "this," "it," and "they" are the most common source of genuine ambiguity in student papers. Name the thing: "This divergence shows…"
• Find the buried subject. "There was a failure of coordination among the allies" hides who did what. "The allies failed to coordinate" is shorter, clearer, and assigns responsibility — which in political science is usually the whole point.
• Break up sentences over about 35 words. Long sentences are not more sophisticated; they are just harder to follow. If a sentence has three clauses stacked on each other, it is probably two sentences.
• Cut the throat-clearing. "It is important to note that," "in today's society," "throughout history." Delete them and check whether anything was lost. Usually nothing was.
• Check that each sentence connects to the one before it. Clarity often fails not within sentences but between them — the logical step is in your head and not on the page.
• Watch for the terms you have not defined. If a key concept first appears on page four doing important work, the reader needed it defined on page one.
• Prefer the plain word. "Utilise" is not better than "use." Writing that sounds like it is trying to sound academic reads as insecure; writing that says a complicated thing simply reads as confident.

A test worth applying at the end: could an intelligent person who has not taken this course follow your argument? Not every technical detail — but the question, the claim, and why the evidence supports it. If you have a friend or a roommate willing to read it, use them. If not, leave it a day and read it as a stranger.

One practical note on formatting, since it belongs to the same principle: follow the assignment's requirements on font, spacing, margins, and length. Double-spacing exists so that I can write between the lines. A paper that is hard to annotate gets less useful feedback, which costs you on the next one.

Revision checklist

Print it or read it on a screen you have not been writing on. Then:

1. Underline your thesis. If you cannot find one sentence to underline, that is the problem.
2. Read the first sentence of each paragraph in sequence. Does it track?
3. Read your headings as a list. Do they describe a coherent progression from question to answer?
4. Scan for paragraphs longer than half a page. Split each one at the point where the job changes.
5. For every paragraph, ask: what claim does this establish, and how does it advance the thesis? Cut or rewrite anything with no answer.
6. Check that every factual claim a reader could dispute has a citation, and that every citation appears in your reference list.
7. Check every quotation against the original — wording, and whether the author actually meant what you are using them to mean.
8. Have you stated and answered the strongest objection?
9. Read it aloud. You will hear the sentences that do not work.
10. Check the mechanics: formatting, word count, file name, and the specific requirements in the assignment sheet.

Time

A paper written the night before reads like a paper written the night before, and the tell is always the same: the argument arrives in the final paragraph, because that is when the writer figured it out. Writing is how you discover what you think. Budget for a draft you will substantially rewrite, and leave at least a day between finishing and revising so you can read your own prose as a stranger would.